Willingness-to-Pay for District Heating from Renewables of Private Households in Germany

Page created by Theresa Riley
 
CONTINUE READING
sustainability

Article
Willingness-to-Pay for District Heating
from Renewables of Private Households in Germany
Thomas Krikser 1, *, Adriano Profeta 2 , Sebastian Grimm 3 and Heiko Huther 3
 1    Department of Management in the International Food Industry, University of Kassel, 34125 Kassel, Germany
 2    DIL—Deutsches Institut für Lebensmitteltechnik e.V., 49610 Quakenbrück, Germany; a.profeta@dil-ev.de
 3    AGFW | DerEnergieeffizienzverband für Wärme, Kälte und KWK e.V., 60596 Frankfurt am Main, Germany;
      s.grimm@agfw.de (S.G.); h.huther@agfw.de (H.H.)
 *    Correspondence: t.krikser@uni-kassel.de
                                                                                                      
 Received: 24 April 2020; Accepted: 15 May 2020; Published: 18 May 2020                               

 Abstract: Recent discussion about future energy production promotes the recovery of industrial heat
 as a potential solution to reduce carbon emissions. Experts call for an expansion of central heating by
 renewable energy systems to ensure a decarbonization of energy systems. In this context, district
 heating could play a significant role in city and district planning. Nonetheless, for private households,
 environmental aspects are only one factor amongst others as, e.g., capital costs, comfort and security.
 In this study, we focus on private household preferences and willingness-to-pay for district heating
 and district heating from renewables compared to gas condensing boilers and heat pumps. For the
 study, we decided to apply a discrete-choice experiment and collected data on attitudes towards
 sustainability, economic aspects and demands for providers of heat supply as dimensions for a factor
 and cluster analysis in order to apply a market segmentation. The results show that district heating
 by renewables is the most preferred heating option for households followed by district heating from
 fossil fuels, heat pumps and gas boilers. Furthermore, the study offers more profound insight into
 the willingness-to-pay for each heating option and reports interaction effects for the different market
 segments that could be identified in the analysis.

 Keywords: district heating; renewables; household preferences; discrete-choice; WTP-space

1. Introduction
     The heating system is an essential component of the maintenance and improvement of comfort in
a home, as heat energy is required for heating rooms, incoming air and domestic hot water. Space
heating demand accounts for a significant fraction of the overall residential energy demand. Private
households in Germany consumed about 1570 PJ for heating in 2015; this means more than 69% of
the total final energy demand for residential purposes [1]. In Germany, residential heat supply is
mainly based on two fossil fuels, heating oil and natural gas. According to the German federal ministry
for economic affairs and energy [2], in 2016 about 47% of existing homes were using gas for heating,
while about 24%, had an oil-based residential heating system (RHS). Moreover, 44% of the newly built
homes were equipped with a gas-fired RHS [3].
     Due to this vital role of oil and gas, residential heating is strongly linked to policy considerations
related to global warming, the security of energy supplies and increasing energy prices [4].
Several studies emphasized the potential for renovations to reduce greenhouse gas emissions (GHG) [5].
Moreover, the expansion of district heating is an essential option for becoming independent of fossil
fuels [6]. Studies show that in the future district heating will be more cost-efficient, and especially in
large cities the capital costs will become rather low [7,8]. District Heating offers the possibility to use

Sustainability 2020, 12, 4129; doi:10.3390/su12104129                       www.mdpi.com/journal/sustainability
Sustainability 2020, 12, 4129                                                                        2 of 14

a variety of heat sources. On the one hand, it can use heat sources that are often wasted; on the other
hand, heat from renewables can be used to produce carbon-neutral energy [9–11].
     From many experts district heating from renewables is seen as one crucial path for decarbonizing
the EU energy system [12–14]. Furthermore, studies show that district heating produced by
a biomass-based combination of heat and power (CHP) is the most widely acceptable heating
alternative in some areas [15–17]. Nonetheless, the renovation of building stocks, optimized isolations
and passive houses can lead to new energy efficiency and therefore to a lower energy demand that can
become challenging to district heating providers [18,19]. In this scenario, a gradual expansion of district
heating to private households could be a future solution. For this purpose, it requires a long-term
planning process that should include costs, resources and environmental and climate aspects [12,20].
In Germany, district heating is becoming more and more important [21,22]. While in the year 2000 only
7% of all new constructed private buildings were supplied by district heating, this value increased to
24% in 2016 [23].
     Policy goals for GHG emission reductions and energy efficiency are lacking new policy measures
in building new houses and renovating existing houses. For this purpose, it is crucial to identify the
key determinants and heterogeneity of these choices. Besides newly constructed housing, in particular,
old houses with oil boilers have a significant potential for the replacement of non-renewable fuel
with some alternative heating system or fuel with lower GHG emissions [24]. In 2017, 0.49 million
Germans indicated they would renovate their private heating systems within the next two years.
Another 1.26 million are thinking of renovating, but the final decision has not yet been made [25].
Therefore, existing houses represent a significant potential source of emission reductions as well,
compared to new house construction.
     This paper is focused on private households’ preferences for district heating and district heating
from renewables. Both homeowners and renters are included. We applied a Discrete-Choice Experiment
(DCE) to compare the preferences and willingness-to-pay for different heating alternatives (district
heating from fossil fuels, district heating from renewables, heat pump and gas condensing boiler).
The DCE-method allows for a detailed analysis of household preferences for the different heating
alternatives. From the results, the additional willingness-to-pay a price premium for district heating
from renewables is derived.
     In previous studies employing choice modelling, the focus was on customer, home and spatial
characteristics, whereas heating system attributes have not been considered. Furthermore, these
studies mostly relied on ownership data; in such cases, the choice is understood as ownership or
availability rather than the actual or planned purchase of a heating system. For example, Vaage (2000)
and Braun (2010) analyzed the determinants of (actually) owning the heating system, using Norwegian
and German cross-sectional household data, respectively [26,27].
     Our study extends previous works employing choice modelling by explicitly presenting
environmental impacts for each labelled heating system as absolute emission levels and customizing
the choice sets according to the availability of district heat. Utilizing our choice modelling results,
we then aim to calculate the willingness-to-pay for different heating systems. In order to analyze the
Willingness-to-Pay (WTP) for different household and market segments, a cluster analysis was carried
out based on the attitudes of the households towards local providers, sustainability and economics of
heating systems. The final clusters were connected to the WTP. Furthermore, we considered, concerning
WTP changes, if the household is in the process of planning a renovation of the heating system.

2. Materials and Methods

2.1. Questionnaire and Sampling
     The objective of this study is to analyze preferences of individuals who are making or planning
heating system decisions. Correspondingly, the study sample comprises individuals who are currently
building or planning a new detached house. Homeowners who are replacing or planning to replace their
Sustainability 2020, 12, 4129                                                                      3 of 14

existing heating system enter the sample as well. A third group are renters who consider the heating
system and the associated costs when they are looking for a new flat (i.e., all adult Germans potentially
belong to this group). The consumer sample of the survey was drawn from an online panel from the
company Respondi. Respondi is an international ISO-certified panel provider. The internationally
valid ISO 26362 standard evaluates the design and the operating effectiveness of quality management.
Four hundred and ninety-two respondents completed the survey. Table 1 presents descriptive statistics
for the sample.

                                   Table 1. Demographics of end-users in Germany.

                                   Characteristic             Sample          Germany
                                      Gender                 (n = 482)     (Genesis 2018)
                                       Male                   49.2%            49.35%
                                      Female                  50.8%            50.65%
                                      Income
                                      < 1,300€                21.1%            22.9%
                                    1,300–2,600€              37.0%            39.2%
                                    2,600–3,600€              23.5%            17.6%
                                    3,600–5000€               12.5%            12.6%
                                       > 5000€                 5.9%             7.8%
                                        Age
                                       18–30                  22.4%            18,4%
                                       31–45                  26.6%            21,9%
                                       46–65                  42.9%            35,6%
                                        > 65                   8.1%            24,1%
                                 Place of residence
                              Small Town (pop.  100,000), centre     18.7%
                                                                                31.9
                           Big City (> 100,000), periphery    22.4%

     The data show that the sociodemographic of the sample is quite similar to the German population.
Nonetheless, within the sample, the generations above 45 years old are underrepresented, and people
in big cities are overrepresented.
     Besides sociodemographic data, the questionnaire consists of data about households’ experiences
with different forms of heating systems, images of different heating systems, attitudes about providers,
price and sustainability of heating systems, scenarios about the future and discrete choice sets with
varying prices and different selections of heating systems.

2.2. Discrete-Choice Experiment
     In addition to questions on socio-demographic characteristics and agreements with statements on
energy and heating systems, a DCE for heating systems was included in the questionnaire. After the
DCE, respondents were asked about their current heating system, whether they had replaced the
original heating system in the last years and whether they had any plans to replace or renovate the
current primary heating system in the future.
     The DCE method is based on microeconomic theory according to which individuals always try to
maximize their benefit [28].
     In a choice experiment, respondents indicate their preference through (repeated) decisions, with
which they choose the one alternative out of different alternatives or goods (evoked set) that they
prefer most. The choice can be interpreted as the discrete variable that possesses the characteristics
‘chosen’ or ‘not-chosen’. According to Lancaster (1966), a good or product can be understood as
Sustainability 2020, 12, 4129                                                                                              4 of 14

a bundle of different objective characteristics from which a consumer derives utility for this good [29].
The Nobel price winner McFadden (1974) refined this approach with recourse to the random utility
theory of Thurstone (1927) [28,30]. Hereby, it was possible to calculate the utility that consumers derive
from the different product characteristics based on choice decisions via logistic regression models.
The DCE method has the advantage of accounting for the multi-attribute nature that characterises
most behavioural decision-making processes.
     As mentioned, in DCEs, individuals must choose from a set of different products offered at
determined prices. The products differ regarding the tested product attributes (e.g., annual operation
costs, etc.). According to microeconomic theory, participants will choose the product with the highest
benefit. Employing DCEs, individuals’ benefit for each tested product attribute can be measured, as
well as the influence of each product attribute on the probability of purchasing the product.
     On the basis of the attributes in Table 2, an experimental design plan was developed for carrying
out a discrete-choice experiment. A choice design consists of sets of several alternatives, and each set of
alternatives is called a choice set. Each single alternative can be described by the different attributes and
attribute levels. The chosen attribute levels are based on literature research and on expert judgment.
The final choice task for the respondent is to choose the alternative she or he prefers most out of the
given set.

                   Table 2. Attribute and characteristics in the Discrete-Choice Experiment (DCE).

                Attribute                   Characteristic 1       Characteristic 2      Characteristic 3   Characteristic 4
                                          District heating from    District heating          Ground               Gas
             Heating system
                                               fossil fuels       from renewables        heat/heat pump     condensing boiler
        Annual operation costs/m2                  6€                    8€                    10 €               12 €
          Investment costs in € *                  4k                    6k                    8k                 10 k
          Primary energy factor                    0.0                   0.7                   0.8                 1.1
      CO2 -emission in kg CO2 /   m2 /a            0.4                   13                    14                  20
                Price risk                        low                  middle                 high
                                             * (without 3 k € for ventilation system).

     The DCE of heating systems was conducted using choice sets offering three heating system
alternatives with varying level attributes: (1) Annual operating cost, (2) investment cost, (3) primary
energy factor, (4) CO2 emissions and (5) price risk (Table 2). The heating systems in the choice sets
were (1) district heating from fossil fuels, (2) district heating from renewables, (3) ground heat/heat
pump and (4) gas condensing boiler. The investment costs had the levels 4000 €, 6000 €, 8000 € and
10,000 € for private households.
     The respondents were asked to choose for each choice set the heating option they would choose if
they were to buy/construct a new house or renovate their actual systems, and if there were no other
options available. Moreover, they were told that they should assume that their house or flat was
currently either located near a heating network or connected to such a network. The investment costs
were related to a house/flat measuring 120 m2 .
     The choice of attributes and their levels (Table 2) was based on earlier studies, on feedback from
experts and on a pre-test of the questionnaire. The same levels of investment cost and required
own work for both house size classes were chosen, as these attributes vary significantly. The annual
operating cost and CO2 and fine particle emissions were calculated based on the energy consumption
of an average detached house, the efficiency of a heating system and the unit price/emission of a fuel
and expressed as annual totals in the choice experiments.
     In total, 12 different (unlabeled) choice sets were generated using the Software NGene [31],
allowing for the main effects to be identified and estimated without confounding. For that purpose,
all parameters mentioned in Table 2 were used to set up an efficient design by using priors based on
expert judgement (e.g., negative priors for cost parameters).
Sustainability 2020, 12, 4129                                                                                                 5 of 14

     Each choice set consisted of three alternatives and each respondent had to make all 12 choice
decisions. Nonsensical choices were eliminated by using restrictions in the generating process of
the experimental design. The choice sets had different orderings of the heating systems to eliminate
possible ordering effects in choices. See Table 3 for a choice set example.

                                                 Table 3. Choice set example.

                                                      Alternative 1                  Alternative 2            Alternative 3
                  Characteristic                District heating renewables   District heating fossil fuels   Heat pump
            Annual operation     costs/m2                  12 €                           12 €                     6€
                 Investment costs                         6000 €                         6000 €                  8000 €
              Primary energy factor                         0,0                           0,7                      0,8
        CO2 emissions in kg CO2     /m2 /year               0,4                            14                      13
                    Price risk                             low                            low                     high

     Our model is estimated in WTP space using a DCE that accounts for unobserved heterogeneity in
correlated WTP coefficients and observable user group and demographic (homeowners vs renters)
characteristics. Train and Weeks [32] recommend this approach and reparametrize the random
parameter (mixed) logit model (RPL) by defining the distribution of WTP directly. Thus a generalized
multinomial logit model (GMNL) is applied [33]. Interaction terms for WTP measures and user
group and demographic variables are included in it. Consequently, significant differences in monetary
valuations of product traits can also be explained based on observable respondent-specific attributes.
The calculated WTPs relate to the annual operation costs per m2 and year because this parameter was
used for GMNL estimation for the monetary evaluation. In the Supplementary Materials a detailed
overview about the applied model is given.
     As a further important feature, unlike most other WTP space studies, we allow for correlations
between the random WTP coefficients of different attributes.

2.3. Cluster Analysis
      During the survey, attitudinal statements were given to the respondents on a five-point rating
scale from “important (1)” to “unimportant (5)”. Overall, 13 variables were used to learn about the
attitudes of the consumers towards local providers, sustainability and economics of heating systems.
For market segmentation of the households, these 13 variables were reduced to a lower number of latent
variables, utilizing principle component analysis. The aim is to display the high quantity of variables
while simultaneously obtaining a low loss of variance. The final number of factors was determined
by the Kaiser criterion (Eigenvalue > 1) and crosschecked by plotting a scree plot [34]. For a more
straightforward interpretation of the factors, varimax rotation was applied. Finally, Cronbach’s alpha
scores were used as a measure for internal scale reliability.
      In order to achieve a customer segmentation, k-means clustering was applied to the factors.
Cluster analysis was used for the minimization of variance within different groups of costumers in the
dataset. For calculating the distance between the individuals, squared Euclidean distance was used.
For the determination of the final number of clusters, the Ward method was performed beforehand.
Additionally, outliers were eliminated by single linkage clustering. Initially, the dendrogram and the
elbow criterion were observed to determine the final number of clusters.

3. Results

3.1. Cluster Analysis
     In a first step, the analyses focused on participants’ expectations about energy suppliers.
These expectations were analyzed by 13 items concerning price, transparency, form of partnership and
environmental impact (see Table 4). For each item, a five-point-rating scale was used with a range from
Sustainability 2020, 12, 4129                                                                                     6 of 14

important to unimportant. A principal component factor analysis was conducted on the 13 items with
varimax rotation. The Kaiser–Meyer–Olkin (KMO) measure verified the sampling adequacy for the
analysis, KMO = 0.904, and all KMO values for individual items were higher than 0.84, which is well
above the acceptable limit of 0.5 [35]. As the criterion for extracting the final factor solution, an initial
analysis was run to obtain eigenvalues for each factor in the data. Three factors had eigenvalues over
Kaiser’s criterion of 1 and in combination explained 67.11% of the variance.

      Table 4. Factor loadings from principal component factor analysis: Communalities, eigenvalues,
      percentages of variance, and alpha value for items of preferences of heating systems (N = 492),
      KMO = 0.904.

                                                                 Factor Loading
                                                  1
                                                                         2                   3
                    Item                      Price and                                                 Communality
                                                                  Local Provider       Sustainability
                                          Financial Security
    Reasonably priced heating system              0.84                 0.11                 0.10           0.72
     Reasonably priced initial outlay             0.78                 0.20                 0.12           0.66
        High security of supply                   0.75                 0.13                 0.33           0.69
       Low fluctuations in prices                 0.70                 0.35                 0.10           0.61
         Transparent accounting                   0.70                 0.20                 0.31           0.63
          High trust in provider                  0.52                 0.45                 0.38           0.61
          Local contact person                    0.25                 0.82                 0.04           0.73
              Local provider                      0.01                 0.75                 0.33           0.68
      Direct contact to the provider              0.33                 0.70                 0.17           0.62
         Long-term partnership                    0.28                 0.68                 0.23           0.55
              Low pollution                       0.17                 0.21                 0.86           0.80
     High share of renewable Energy               0.18                 0.23                 0.83           0.77
    Future efficiency of energy source            0.46                 0.21                 0.62           0.64
                Eigenvalue                       6.13                  1.46                 1.14
          Variance explained in %                47.13                 11.20                8.78
                     α                           0.83                  0.78                 0.79
                                    Note. Boldface indicates the highest factor loadings.

      Table 4 shows the factor loadings after varimax rotation. The items that cluster on the same factor
suggest that factor 1 represents a preference for price and financial security (F1), factor 2 represents
a preference for local providers (F2) and factor 3 represents responsibility for the environment and
sustainability (F3). Cronbach’s α measured the reliability of the factors. All factors had high reliability;
all Cronbach’ α were above 0.78, which is considered “good” regarding internal scale reliability.
      Factor 1 combines positive attitudes towards price and financial security. The variables reflecting
these factors mainly include reasonable pricing, transparency and stable prices. Further variables of
this factor are high trust in the provider and high security of the supply. The second factor identifies
mainly locally attitudes towards the important characteristics of heating systems. Direct contact to
a local provider and a long-term partnership belong to factor 2. F3 pools items which display the
importance of sustainability (e.g., pollution of a high share of renewable energies).
      Based on the three factors, three clusters were identified after eliminating six outliers (see Figure 1).
The results confirm the existence of different consumer segments regarding attitudes towards heating
systems. The first and largest cluster, “The Locals”, represents 56.3% of the sample. These consumers
show less sensibility for prices and prefer a local provider. At the same time, they show some
preferences for sustainability.
      The second cluster, “The Price Sensitives”, is characterized by high scores for the factor “price and
financial security”; the factors “local provider” and “sustainability” are not as relevant as in the other
clusters. The third cluster “The Ecologists” consists of people who take care of environmental aspects.
People in this cluster exhibit a relatively high sense of responsibility for environmental issues, but do
not care about a local provider.
Sustainability  2019,12,
 Sustainability2020,  11,4129
                          x FOR PEER REVIEW                                                                               7 of
                                                                                                                            of 14
                                                                                                                                14

                        Figure 1. Cluster centres of the k-means cluster analysis in Germany.
                        Figure 1. Cluster centres of the k-means cluster analysis in Germany.
3.2. Discrete-Choice Experiment
 3.2. Discrete-Choice Experiment
      The results in Table 5 demonstrate that district heating from renewables is the most preferred
heatingThe   results
          system    in in Table
                       this study5 based
                                    demonstrate     that district
                                            on the annual           heating
                                                            operation   costsfrom    renewables
                                                                               and under           is the most
                                                                                           the assumption      of preferred
                                                                                                                   identical
 heating system
investment      costs in
                       forthis
                           the study
                                heatingbased     on the
                                           systems       annual
                                                     under          operation The
                                                             consideration.     costshouseholds
                                                                                        and underindicated
                                                                                                      the assumption
                                                                                                                  a WTP of    of
10.32  € perinvestment
 identical     m2 /a for this
                            costs  for the heating
                               alternative,           systems
                                               compared    to theunder   consideration.
                                                                   reference  gas condenserThe households       indicated
                                                                                                boiler. In second         placea
 WTP ofdistrict
follows    10.32 € per    m2/afrom
                     heating    for this alternative,
                                      fossil           compared
                                              fuels with  a WTP of  to5.43  € per m2 /a.
                                                                       the reference   gasCompared
                                                                                           condenser to   boiler.  In second
                                                                                                            the reference,
 place follows
ground    heatingdistrict     heatingand
                     had a positive      from   fossil fuels
                                             significant  WTP with   a WTP of 5.43 € per m2/a. Compared to the
                                                                 as well.
 reference,    ground heating
      The parameter       for thehad    a positive
                                    investment       andrevealed
                                                  costs   significant   WTP
                                                                     that for as   increase of these costs of 10,000 €,
                                                                              anwell.
       The parameter
the households        mustforbethe   investment costs
                                  compensated      with revealed
                                                          decreasedthat    for anoperation
                                                                        annual      increase of  these
                                                                                              costs   of costs  €/m
                                                                                                          3.45 of      2 /a for
                                                                                                                     10,000   €,
 the households
evaluating     both as must   be compensated
                          equivalent.               with or
                                           For a house     decreased
                                                              flat withannual    2
                                                                         120 m operation       costs of physical
                                                                                    and an assumed         3.45 €/mlife 2/a for
                                                                                                                             of
 evaluating
20  years forboth     as equivalent.
                 all components       ofFor
                                          theaheating
                                               house orsystem,
                                                         flat withthis
                                                                    120compensation
                                                                         m2 and an assumedadds physical
                                                                                                 up to 8,280life €of(120    m2
                                                                                                                      20 years
×for
   20a     3.45€/m /a). of
      all×components 2      Thetheuniform
                                    heatinglifetime
                                               system,was
                                                        this assumed
                                                              compensation      addscomparability,
                                                                         for better     up to 8,280 € and (120itmis known
                                                                                                                    2 x 20a x
 3.45€/m
that  some 2/a). The uniform lifetime was assumed for better comparability, and it is known that some
              individual components of the heating system have a significantly longer lifetime. These
 individual components
investment      costs are more  of critical
                                   the heating    system
                                             for the       have a choice
                                                     household      significantly
                                                                            of the longer
                                                                                     heatinglifetime.
                                                                                              system These
                                                                                                        comparedinvestment
                                                                                                                        to the
 costs are
annual        more critical
          operation    costs. for the household choice of the heating system compared to the annual
 operation costs.
                           Table 5. Results estimation in Willingness-to-Pay (WTP) space.
                            Table 5. Results estimation in Willingness-to-Pay2 (WTP) space.
                                                                       WTP in €/m /a
                     Parameters and Interactions                                       Std. Error              z-Value
                                                                         (coeff.)
                                                                             WTP in €/m2/a
                      Parameters and interactions                                                     Std. error       z-value
                 mean estimates random parameters                                  (coeff.)
                   district heating from fossil fuels                     5.43 ***             0.46             11.89
                 mean   estimates random parameters
                   district heating from renewables                       10.32 ***            0.44             23.45
                   district  heating
                        ground         from
                                 heat/heat   fossil fuels
                                           pump                            3.29 *** 5.43 ***   0.32     0.46    10.29 11.89
                   district  heatingcosts
                      investments     from   renewables
                                          * 10.000                        −3.45 ***10.32 ***   0.48     0.44    −7.13 23.45
               sd. meanground
                          estimates  random pump
                                  heat/heat   parameters                           3.29 ***             0.32            10.29
                   district heating from fossil fuels                      3.03 ***            0.32                9.49
                       investments costs * 10.000                                  –3.45 ***            0.48             –7.13
                   district heating from renewables                        3.29 ***            0.23                14.23
               sd. mean    estimates
                        ground         random
                                 heat/heat pumpparameters                  2.10 ***            0.24                 8.89
                      investments
                   district  heatingcosts * 10.000
                                       from  fossil fuels                  2.08 *** 3.03 ***   0.56     0.32       3.69 9.49
                   district heating from renewables                                3.29 ***             0.23            14.23
                         ground heat/heat pump                                     2.10 ***             0.24            8.89
                       investments costs * 10.000                                  2.08 ***             0.56            3.69
                   mean estimates interaction ‘ecos’
                district heating from fossil fuels x ecos                         –1.71 **              0.59            –2.91
                district heating from renewables x ecos                           1.41 ***              0.37            3.78
                     ground heat/heat pump x ecos                                   0.26                0.34             0.77
Sustainability 2020, 12, 4129                                                                                             8 of 14

                                                            Table 5. Cont.

                                                                             WTP in €/m2 /a
                       Parameters and Interactions                                                 Std. Error   z-Value
                                                                               (coeff.)
                     mean estimates interaction ‘ecos’
                  district heating from fossil fuels × ecos                      −1.71 **             0.59       −2.91
                  district heating from renewables × ecos                        1.41 ***             0.37       3.78
                       ground heat/heat pump × ecos                                0.26               0.34        0.77
              mean estimates interaction ‘price sensitives’
            district heating from fossil fuels × price sensitives               −3.03 ***             0.68       −4.43
           district heating from renewables × price sensitives                  −3.34 ***             0.36       −9.18
                ground heat/heat pump × price sensitives                        −1.03 **              0.37       −2.79
                    mean estimates interaction ‘locals’
                  district heating from fossil fuels × locals                    −0.37                0.31       −0.79
                 district heating from renewables × locals                        0.31                0.47        0.93
                      ground heat/heat pump × locals                            −1.05 ***             0.31       −3.39
                mean estimates interaction ‘house owner’
             district heating from fossil fuels × house owner                   −2.38 ***             0.47       −5.07
             district heating from renewables × house owner                     −2.53 ***             0.28       −8.92
                  ground heat/heat pump × house owner                           −0.73 **              0.26       −2.82
                     mean estimates interaction ‘plan’
                  district heating from fossil fuels × plan                       1.58 **             0.60       2.63
                  district heating from renewables × plan                         −0.38               0.37       −1.04
                       ground heat/heat pump × plan                              −1.17 **             0.36       −3.26
           sd. mean estimates correlated random parameters
         district heating from fossil fuels/district heating from r.             1.02 **              0.32       3.23
               district heating from fossil fuels/ground heat                   0.99 ***              0.22       4.59
            district heating from fossil fuels/investment costs                 −6.10 ***             0.42      −14.53
                    district heating from r./ground heat                        3.26 ***              0.24      13.83
                  district heating from r./investment costs                     −6.27 ***             0.32      −19.46
                       ground heat./investment costs                              0.75                3.11       0.24
                                 scale (mean)                                   −3.83 ***             0.08       -5.12
                                scale heter (τ)                                 0.90 ***              0.07       14.02
                                 observations                                     5,784
                                Log likelihood                                  −4,077.20
                                Note: ***, **, * significant at the 0.1%, 1% and 5% levels, respectively.

      If the fact that investment costs for access to district heating are in general lower in comparison to
investment costs for gas condensing boiler is accounted for, the WTP for district heating is reduced;
e.g. if the difference is 10.000 €, the calculated 3.45 € must be subtracted (WTP-district-heating fossil
fuels reduced = 5.43 € – 3.45 € = 1.98 €). In Germany in 2018, the cost difference between district
heating and gas condenser boiler was on average about 2.30 € per m2 /a [36]. Therefore, the revealed
WTP in the survey for district heating from fossil fuels reduced by the difference in investment costs is
about the cost difference that can be found in the market. Nevertheless, full cost accountings for gas
boilers and district heating show smaller cost difference between the heating systems, especially for
multi-family housing [37].
      It should be highlighted that in this study households stated a significant additional price premium
just for the fact that the district heating is coming from renewables instead of fossil fuels. However,
it has to be considered that households could have overstated their WTP because they were only
making a hypothetical choice. Despite this fact, the results are still a good indicator for the preference
ranking of the different heating system. All standard deviations of the random parameter coefficients
in the GMNL model are significantly different from zero. Respondents weight the attributes differently
and averaging these weights across respondents leads to statistical insignificance.

3.2.1. Interaction Effects
    Concerning the different clusters, a high degree of heterogeneity between the segments can be
found, as can be seen from the numerous significant interactions. It appears plausible that Ecologists
revealed a significantly higher WTP for district heating from renewables (∆+1.41 €), whereas they
Sustainability 2020, 12, 4129                                                                                     9 of 14

pay less for district heating from fossil fuels (∆–1.71 €). This finding is in line with research for green
electricit; e.g., Hille et al. [38] found that ecologically orientated people are willing to pay a price
premium for electricity produced from renewable energy sources.
     Price sensitives show a lower WTP for district heating from fossil fuels (∆–3.03 €) and district
heating from renewables (∆–3.34€). For this group, it can be hypothesized that these do not want to
be bound for a long time on only one system so that they can switch to the cheapest heating option
or cheapest supplier. Interestingly, house owners had a below-average WTP for each kind of district
heating (∆WTPfossil: –2.38 €, ∆WTPrenewables: –2.53 €). The reasons for that behaviour are unclear.
It could hypothesize that house owners have a higher need to be free in their energy supplier choice or
that they are satisfied with the current heating system.
     Respondents in the process of planning a new heating system revealed an above-average WTP for
district heating from fossil fuels (∆+1.58 €) and the WTP for district heating from renewables is not
reduced in comparison to the rest of the sample. Therefore, it can be stated that even when it comes to
the real-world decision, district heating is still highly preferred.
     It should be noted that, concerning calculation of the WTP, for one of the mentioned segments,
the interaction effects have to sum up to the main effect; e.g., for the price sensitives and district heating
from renewables, the calculation is:

                WTP for district heating from renewables for the segment price sensitives =

               WTP district heating from renewables + WTP district heating from renewables × price sensitives

     For the mentioned case, the WTP for district heating from renewables for the segment price
sensitives is 6.98 € (10.32 €–3.34 €) and 2.40 € (5.43 €–3.03 €) for district heating from fossil fuels.

3.2.2. Interactions between Attribute Preferences
      The random parameters are defined as being correlated with each other. Table 6 presents the full
correlation matrix of the random parameters for the GMNL-WTP-space model. There is a statistically
significant positive association between the preference for district heating from fossil fuels and district
heating from renewables, i.e., those respondents who exhibit a higher WTP district heating from fossil
fuels are more likely to value district heating from renewable alternative as well. This correlation is
less strong between ground heating and the district heating options.

                                   Table 6. Correlation matrix random parameters.

                                   d.c. from          d.c. from
             Parameters                                                   Ground Heat         Investments Costs
                                  Fossil Fuels       Renewables
           d.c. from fossil
                                      1.00                0.85                 0.44                   −0.89
                 fuels
              d.c. from
                                      0.85                1.00                 0.58                   −0.89
             renewables
            ground heat               0.44                0.58                1.00                    −0.24
         investments costs           −0.89               −0.89                −0.24                    1.00

3.2.3. Scale Heterogeneity
     Besides preference heterogeneity, we also find statistically significant scale heterogeneity. Therefore,
the assumption of identical scales across individuals should be rejected. The value of the parameter
of scale heterogeneity (0.639) showed a considerable scale heterogeneity for respondents (Table 5).
The rather quantitatively tremendous value implies that a certain degree of randomness existed in their
choice decisions, and hence a certain degree of uncertainty. A plausible explanation for this specific
degree of uncertainty was that many respondents might not be able to disentangle sophisticated
Sustainability 2020, 12, 4129                                                                        10 of 14

ecological and environmental attributes and discern the utility changes associated with different
heating options.

4. Discussion
      The results clearly show that ‘district heating from renewable energies’ is the most preferred
heating option for the analyzed households. The ranking of the other alternatives is ‘district heating
from fossil fuels’, ‘heat pump’ and ‘gas’. It should be highlighted that the participants revealed
a significant additional WTP for ‘district heating’ just for the fact that it is from renewable energies.
The WTP is about 5 € higher for this option than for ‘district heating from fossil fuels’. Nevertheless,
the WTP differs between different consumer segments in the market. During the study, we could
identify more than 18% of the participants in our survey as price sensitive. Within this group, the WTP
for district heating from renewables but also district heating from fossil fuels is much lower than for
the average participant in the survey. On the other hand, we can see a segment of more than 25% that
has an above average WTP. We call this segment “The Ecologists”; this cluster of people stated a high
preference for low pollution and a high share of renewable energy. Furthermore, the cluster we called
“The Locals” also has a high WTP for district heating, especially for district heating from renewables.
“The Locals” represents about half of the sample.
      Even if we reconsider that the probability of overestimating the WTP is very high in a discrete choice
experiment, the data show significant preferences for district heating and renewable energy systems.
Another issue is the representativeness of the sample. In our sample we have an overrepresentation of
people age 45 and younger and an underrepresentation of people age 45 and older. This may affect
the results in both ways; on one hand, older people do usually have a better financial background;
on the other hand, younger people often have a more positive attitude towards ecofriendly technology.
Nevertheless, the findings demonstrate that house owners and renters in Germany are aware of the
association of energy and global warming and are willing to pay a higher price for ecofriendly energy.
Thus, from a demand side, most parts of the population do already favour district heating. Therefore the
discussions about using district heating for decarbonization [6,9,10] seem to be supported by the
German population. Especially people who plan to build a new heating system prefer district heating,
while house owners have a lower WTP for district heating. Considering the review of Werner [24],
which shows that especially old houses with oil boilers have a high potential for reducing greenhouse
gas emissions, the results show the positive attitudes of consumers towards this potential. On the
other hand, house owners, in general, seem to be more sceptical. This could mean that house owners
who do not plan to renew their heating systems are satisfied with their current situation. In general,
it seems to be possible to use district heating for decarbonization from the demand side.
      “The Price Sensitives” show a low WTP for district heating. Based on the current study, we cannot
explain the reasons for this behaviour. It seems obvious that, especially in this cluster, the estimated
installation and operating costs play a major role. It is possible that district heating is seen as
a complex system that deals with long-lasting contracts up to 10 years (maximum contract period in
AVBFernwärmeV §32). This could give rise to fears that this would be expensive in the long term.
However, district heating is characterized by stable prices, which may only be adjusted according to
defined rules (Preisänderungsklausel AVBFernwärmeV §24). The volatile prices of energy sources
such as gas or oil only lead to minor price fluctuations due to the pro-rata consideration. To analyse
the different motives and drivers for the different clusters of consumers, more research is necessary.
      This study was conducted under the assumption that the various heating alternatives are available
to all end customers and that the decision is primarily based on economic and ecological aspects and
personal preference. In reality, however, in addition to the willingness of the customer, the technical
and organizational requirements must also be met. One condition is the existence of a district heating
network to connect to, which is usually possible mainly in urban areas. About 30% of the total German
population live in the 70 largest German cities, and at the same time more than 50% of the total
district heating sales are generated there [39]. This is because district heating has its advantages above
Sustainability 2020, 12, 4129                                                                            11 of 14

all when as many heat consumers as possible can be connected per meter of pipeline, so densely
populated urban areas with apartment buildings are particularly suitable. However, especially in
the case of apartment buildings, the individual cannot determine his or her own type of heat supply
independently of the other occupants/co-owners of the house owner.
      Since the energy supplier must ensure that its heat generation units can meet the customer’s heat
requirements, the expansion of the district heating grid needs to be planned in detail, especially because
the construction of a new production plant and new pipeline is associated with high investment costs.
In order to minimize the entrepreneurial risk, analyses of potential customers are an important part of
the planning process, also because the connection rate has a massive influence on the heat generation
costs of the individual system [40].
      In recent years, the importance of district heating networks for achieving climate targets has been
demonstrated in numerous studies. European legislators have also taken note of this and no longer
concentrate solely on electricity and transport, but recognize that a successful energy turnaround
is not possible without the heat sector. Large unused potential has been identified in the heating
sector. Today, heating and cooling are seen as the key to accelerated decarburization of the energy
system [41]. This has also led to demands for corresponding investments at the political level national
and in Europe [42].
      If today large amounts of heat are still obtained from highly efficient CHP processes, the integration
of renewable energies is particularly important for the future. The 40/40 strategy drawn up by the
German District Heating Association shows that, in combination with the energetic renovation of
buildings, a reduction of CO2 emissions by approximately 80% can be achieved by 2050 (compared
to 1990). This requires innovative systems and solutions, which are predominantly accompanied by
massive investments and thus make it more difficult to achieve economic viability and, above all,
to compete with conventional fossil supply alternatives. Through suitable support programs and
investment incentives, these potentials can be better utilized in the future [43].
      From a technical point of view, numerous renewable alternatives are already available as potential
heat sources (e.g., solar thermal, biomass, geothermal, power to heat, surplus or waste heat from
industrial companies, etc.). The challenge in many places is to adapt existing district heating systems
so that the technical integration of volatile sources (solar radiation can vary greatly) is possible at
the required temperature level. That this is possible has already been demonstrated by successful
lighthouse projects [44] and the steadily growing number of solar thermal heat sources. The thermal
capacity of solar heat in German district heating grew by around 50% in 2019 to a total of 70 MW [45].
The continuing need for research is underlined by ongoing national and international research projects
in the field of renewable energies (e.g., Upgrade DH, ReUseHeat, Solnet 4.0, etc.). It is also an important
part of the 7th German Federal Government’s energy research programme [46].
      At the end we would highlight the fact that in surveys respondents sometimes tend to give
answers that they think are expected or favoured by those conducting the survey, which may lead
to a gap between the stated WTP and actual the WTP. Furthermore, DCE studies have shown that
respondents often ignore one or more attributes when making choices, which may lead to biased WTP
estimations [47,48]. Therefore, it is planned to merge the collected data with data from other sources to
calibrate and to identify an appropriate scaling factor for the found WTPs. Nonetheless, we assume
that the measured preference structure of the household, that is the relative importance of the attributes
to each other, is stable.

Supplementary Materials: Available online: http://www.mdpi.com/2071-1050/12/10/4129/s1.
Author Contributions: Conceptualization, H.H., S.G., T.K. and A.P.; methodology, T.K. and A.P.; validation, A.P.,
S.G. and H.H.; formal analysis, T.K. and A.P.; resources, H.H.; data curation, T.K. and A.P.; writing—original
draft preparation, T.K. and A.P.; writing—review and editing, T.K. and A.P; supervision, S.G. and H.H.; project
administration, H.H. and S.G.; funding acquisition, H.H. All authors have read and agreed to the published
version of the manuscript.
Sustainability 2020, 12, 4129                                                                                  12 of 14

Funding: This research was funded by European Union’s Horizon 2020 research and innovation programme
under grant agreement number 691624.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
RHS              Residential Heating Systems
WTP              Willingness-to-Pay
GHG              Greenhouse Gas
CHP              Combined Heat and Power
DCE              Discrete-Choice Experiment
RPL              Random Parameter Logit Model
GMNL             Generalized Multinominal Logit Model
KMO              Kaiser–Mayer–Olkin Measure

References
1.    AGEB. Anwendungsbilanzen. Stand 10/2016. 2016. Available online: https://www.ag-energiebilanzen.de
      (accessed on 24 April 2020).
2.    Bundesministerium für Wirtschaft und Energie. Energiedaten: Gesamtausgabe. 2018. Available
      online: https://www.bmwi.de/Redaktion/DE/Artikel/Energie/energiedaten-gesamtausgabe.html (accessed
      on 24 September 2019).
3.    BDEW. Beheizungsstruktur im Wohnungsneubau in Deutschland. 2017. Available online: https://www.bd
      ew.de (accessed on 24 April 2020).
4.    European Commission. An EU Strategy on Heating and Cooling; COM (2016)51; The European Comission:
      Brussels, Belgium, 2016.
5.    Eriksson, L.; Morandin, M.; Harvey, S. A feasibility study of improved heat recovery and excess heat export
      at a Swedish chemical complex site. Int. J. Energy Res. 2018, 42, 1580–1593. [CrossRef]
6.    Rezaie, B.; Rosen, M.A. District heating and cooling: Review of technology and potential enhancements.
      Appl. Energy 2012, 93, 2–10. [CrossRef]
7.    Persson, U.; Werner, S. Heat distribution and the future competitiveness of district heating. Appl. Energy
      2011, 88, 568–576. [CrossRef]
8.    Paiho, S.; Reda, F. Towards next generation district heating in Finland. Renew. Sustain. Energy Rev. 2016, 65,
      915–924. [CrossRef]
9.    Sayegh, M.A.; Danielewicz, J.; Nannou, T.; Miniewicz, M.; Jadwiszczak, P.; Piekarska, K.; Jouhara, H. Trends
      of European research and development in district heating technologies. Renew. Sustain. Energy Rev. 2017, 68,
      1183–1192. [CrossRef]
10.   Lake, A.; Rezaie, B.; Beyerlein, S. Review of district heating and cooling systems for a sustainable future.
      Renew. Sustain. Energy Rev. 2017, 67, 417–425. [CrossRef]
11.   Ghafghazi, S.; Sowlati, T.; Sokhansanj, S.; Melin, S. A multicriteria approach to evaluate district heating
      system options. Appl. Energy 2010, 87, 1134–1140. [CrossRef]
12.   Lund, H.; Möller, B.; Mathiesen, B.V.; Dyrelund, A. The role of district heating in future renewable energy
      systems. Energy 2010, 35, 1381–1390. [CrossRef]
13.   Connolly, D.; Lund, H.; Mathiesen, B.V.; Werner, S.; Möller, B.; Persson, U.; Boermans, T.; Trier, D.;
      Østergaard, P.A.; Nielsen, S. Heat Roadmap Europe: Combining district heating with heat savings to
      decarbonise the EU energy system. Energy Policy 2014, 65, 475–489. [CrossRef]
14.   Brocklebank, I.; Styring, P.; Beck, S. Heat mapping for district heating. Energy Procedia 2018, 151, 47–51.
      [CrossRef]
15.   Kontu, K.; Rinne, S.; Olkkonen, V.; Lahdelma, R.; Salminen, P. Multicriteria evaluation of heating choices for
      a new sustainable residential area. Energy Build. 2015, 93, 169–179. [CrossRef]
16.   Kim, H.-J.; Lim, S.-Y.; Yoo, S.-H. Public preferences for district heating system over individual heating system:
      A view from national energy efficiency. Energy Effic. 2019, 12, 723–734. [CrossRef]
Sustainability 2020, 12, 4129                                                                                  13 of 14

17.   Lichtenwoehrer, P.; Erker, S.; Zach, F.; Stoeglehner, G. Future compatibility of district heating in urban
      areas—A case study analysis in the context of integrated spatial and energy planning. Energy Sustain. Soc.
      2019, 9, 5. [CrossRef]
18.   Lygnerud, K. Challenges for business change in district heating. Energy Sustain. Soc. 2018, 8, 10. [CrossRef]
19.   Sameti, M.; Haghighat, F. Optimization approaches in district heating and cooling thermal network.
      Energy Build. 2017, 140, 121–130. [CrossRef]
20.   Erker, S.; Lichtenwoehrer, P.; Zach, F.; Stoeglehner, G. Interdisciplinary decision support model for grid-bound
      heat supply systems in urban areas. Energy Sustain. Soc. 2019, 9, 22. [CrossRef]
21.   Lottner, V.; Schulz, M.E.; Hahne, E. Solar-Assisted District Heating Plants: Status of the German Programme
      Solarthermie-2000. Sol. Energy 2000, 69, 449–459. [CrossRef]
22.   Günther, M.; Greller, M.; Fallahnejad, M. Evaluation of Long-Term Scenarios for Power Generation and
      District Heating at Stadtwerke München. Informatik Spektrum 2015, 38, 97–102. [CrossRef]
23.   BDEW. Erneuerbare Energien und das EEG: Zahlen, Fakten, Grafiken (2016). 2016. Available online:
      https://www.bdew.de/energie/erneuerbare-energien-und-das-eeg-zahlen-fakten-grafiken/ (accessed on
      24 September 2019).
24.   Werner, S. International review of district heating and cooling. Energy 2017, 137, 617–631. [CrossRef]
25.   Arbeitsgemeinschaft Verbrauchs- und Medienanalyse. Konsumenten punktgenau erreichen. Basisinformationen
      für fundierte Mediaentscheidungen. VuMA Touchpoints 2018. Available online: https://www.vuma.de/fileadmin/
      user_upload/PDF/berichtsbaende/VuMA_Berichtsband_2018.pdf (accessed on 24 September 2019).
26.   Vaage, K. Heating technology and energy use: A discrete/continuous choice approach to Norwegian
      household energy demand. Energy Econ. 2000, 22, 649–666. [CrossRef]
27.   Braun, F.G. Determinants of households’ space heating type: A discrete choice analysis for German
      households. Energy Policy 2010, 38, 5493–5503. [CrossRef]
28.   McFadden, D. Conditional logit analysis of qualitative choice behavior. In Frontiers in Econometrics;
      Zarembka, P., Ed.; Academic Press: New York, NY, USA, 1974; pp. 105–142.
29.   Lancaster, K.J. A New Approach to Consumer Theory. J. Political Econ. 1966, 74, 132. [CrossRef]
30.   Thurstone, L.L. A law of comparative judgement. Psychol. Rev. 1927, 34, 273–286. [CrossRef]
31.   ChoiceMetrics. Ngene 1.1.1. User Manual & Reference Guide; ChoiceMetrics: Sydney, Australia, 2012.
32.   Train, K.; Weeks, M. Discrete Choice Models in Preference Space and Willingness-to-Pay Space. In Applications
      of Simulation Methods in Environmental and Resource Economics: The Economics of Non-Market Goods and Resources,
      6th ed.; Scarpa, R., Alberini, A., Eds.; Springer: Dordrecht, The Netherlands, 2005; pp. 1–16.
33.   Fiebig, D.G.; Keane, M.P.; Louviere, J.; Wasi, N. The Generalized Multinomial Logit Model: Accounting for
      Scale and Coefficient Heterogeneity. Mark. Sci. 2010, 29, 393–421. [CrossRef]
34.   Cattell, R.B. The scree test for the number of factors. Multivar. Behav. Res. 1966, 1, 245–276. [CrossRef]
35.   Field, A. Discovering Statistics Using IBM SPSS Statistics. And Sex and Drugs and Rock ’n’ Roll, 4th ed.;
      Sage Publications Inc: Los Angeles, CA, USA, 2013.
36.   CO2online. Heizspiegel Deutschland. 2019. Available online: https://www.heizspiegel.de (accessed on
      24 April 2020).
37.   Fränkle, C. Heizkostenvergleich nach VDI 2067: Musterrechnung. Available online: https://www.agfw
      .de/energiewirtschaft-recht-politik/wirtschaft-und-markt/markt-preise/heizkostenvergleich/ (accessed on
      24 April 2020).
38.   Hille, S.; Weber, S.; Brosch, T. Consumers’ preferences for electricity-saving programs: Evidence from
      a choice-based conjoint study. J. Clean. Prod. 2019, 220, 800–815. [CrossRef]
39.   Blesl, M.; Eikmeier, B. Die 70/70-Strategie: Konzept und Ergebnisse. Available online: https://www.agfw.de/
      energiekonzepte/7070-4040-strategie/ (accessed on 24 April 2020).
40.   AGFW e.V. EnEff:Wärme- Kassel Zum Feldlager. GeosolareNahwärmeversorgung für die Siedlung ‘Zum
      Feldlager’; AGFW Heftreihe Forschung und Entwicklung—Heft; 2018. Universität Kassel: Kassel, Germany.
41.   Moczko, D. Die neue Erneuerbare-Energien-Richtlinie und die Fernwärme.                                  Available
      online: https://www.energie.de/euroheatpower/news-detailansicht/nsctrl/detail/News/die-neue-erneuerb
      are-energien-richtlinie-und-die-fernwaerme-201961/ (accessed on 24 April 2020).
42.   Europäische Kommission. Länderbericht Deutschland 2020. Available online: https://ec.europa.eu/info/sites/
      info/files/2020-european_semester_country-report-germany_de.pdf (accessed on 24 April 2020).
Sustainability 2020, 12, 4129                                                                               14 of 14

43.   Blesl, M.; Koziol, M.; Ludwig, M.C.; Rapp, H.; Tenberg, B.; Vautz, S.; Wolf, S. 40/40-Strategie: Unser
      Konzept für die Wärmewende. Available online: https://www.agfw.de/energiekonzepte/7070-4040-strategie/
      (accessed on 24 April 2020).
44.   Roth, T.; Grimm, S.; Rutz, D. Upgrading the Performance of District Heating Network: Best Practice Examples
      on Upgrading Projects. Available online: https://www.upgrade-dh.eu/images/Publications%20and%20Rep
      orts/D2.1_2019-04-30_Upgrade%20DH_final_AGFW.PDF (accessed on 24 April 2020).
45.   Solar District Heating. Stuttgart. 2020. Available online: https://www.solar-district-heating.eu/de/aktuelles
      /presse/# (accessed on 24 April 2020).
46.   Bundesministerium für Wirtschaft und Energie. Innovationen für die Energiewende: 7. Energieforschungsprogramm
      der Bundesregierung. Available online: https://www.bmwi.de/Redaktion/DE/Publikationen/Energie/7-energieforsc
      hungsprogramm-der-bundesregierung.pdf?__blob=publicationFile&v=14 (accessed on 24 April 2020).
47.   Carlsson, F.; Kataria, M.; Lampi, E. Dealing with Ignored Attributes in Choice Experiments on Valuation of
      Sweden’s Environmental Quality Objectives. Environ. Resource Econ. 2010, 47, 65–89. [CrossRef]
48.   Hensher, D.A.; Rose, J.M.; Greene, W.H. Applied Choice Analysis: A Primer; Cambridge University Press:
      Cambridge, UK, 2005.

                           © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
                           article distributed under the terms and conditions of the Creative Commons Attribution
                           (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
You can also read